The ‘Core’ refers to a selection of about 30,000 pages of serial text on which multi-level processing and editing was to have been carried out. It was developed in the autumn and winter of 2005/6 and it was abandoned by the spring. Nevertheless, the process of the selection of criteria for the core enabled the team to familiarise itself further with the characteristics of the six titles, individually and as a cluster in the edition, to hazard identification of defining elements of serials, and to test the degree to which these were shared by our diverse titles.

In the trajectory of development of ncse the Core was Plan B. It was devised in response to two assumptions that changed. The first was that the project would consist of 30,000 pages of text. The discovery of the multiple editions prompted us to revise this estimate to around 110,000 pages. As this almost tripled the amount of pages that we would have to review, edit, and markup we began to reconsider the amont of manual editorial work it would be possible to do. The second assumption was that the model of multi-level processing envisaged by the team and Olive software, in which segmentation would be carried out on items and the departments in which they were nested, could be perfected and adopted. On the understanding that multi-level processing could only be managed and afforded on 30,000 pages the research team decided to devise a selection process that would provide a manageable sub-section of the resource on which these advanced editorial and computational methods could be applied. Affording was a matter of time for Olive, CCH, and the research team as well as cost.

While the core was to be a diagnostic space on which intensive processing and editing would be demonstrated, and possibilities of electronic publication of serials explored, the entire resource (i.e. all 100,000 pages) would also be released to users in readable and, on a basic level, searchable form so that the Core would be accessible in its serial contexts.

Four criteria were selected for inclusion of resource in the Core. They were:

  • the beginning and ending of each generic title
  • moments of change of editor
  • proportional representation of visual elements in the cluster
  • thematic content: court prosecution for libel and sedition of editors

The resulting tranche was then finessed with respect to further criteria:

  • chronological spread across the century
  • title spread, and some attempt to reflect the proportion of the edition that individual runs represented
  • continuity: in order to minimise fragmentation of the resource and to honour continuity as a basic principle of seriality, we decided that the smallest fragment would be a calendar year.
  • Inclusion of multiple editions: it was decided to represent multiples at moments of overlap between titles. 1837-8 (Monthly Repository and Northern Star) and 1850-2 (Northern Star and Leader)

The Core was abandoned in the spring of 2006 for several reasons. First, the research team and Olive came to the conclusion that multi-level processing was too difficult to achieve at acceptable standards within the time and budget at our disposal. Instead, we worked out a way to deliver the characteristics of dual level processing (i.e. items and departments) on the basis of single level processing and metadata entry. In the summer of 2006 Olive offered to process the whole 100,000 pages to this specification. We needed to decide whether we would go for the smaller processed core, on which we could do more intensive editing or a larger more consistent resource on which all parties could do less. Deliberations among the research team and the larger project team took place, and an assessment of automatic processes of text mining, entity extraction, and semantic tagging was made. The other criterion was the interest of our users and the larger world of electronic resources of nineteenth-century print. This combination of criteria led us to abandon the Core, and go for uniform processing across 100,000 pages.

We developed various graphic representations of the Core, which are available here; there is also a fuller explanation of the Core in a research paper available here (pdf).